BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//ZContent.net//ZapCalLib 1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VEVENT
DTSTART;TZID=Atlantic/Canary:20190628T103000
DTEND;TZID=Atlantic/Canary:20190628T113000
UID:iactalks-1309
X-WR-CALNAME: IAC Talks: Open Astronomy Seminars
X-ORIGINAL-URL: /iactalks/Talks/view/1309
CREATED:2019-06-28T10:30:00+01:00
X-WR-CALDESC: IAC Talks upcomming talks
SUMMARY:MaxiMask: A new tool to identify contaminants in astronomical image
 s using convolutional neural networks.
DESCRIPTION:MaxiMask: A new tool to identify contaminants in astronomical i
 mages using convolutional neural networks.\nMaxime Paillassa\n\nWe propose
  to use convolutional neural networks to detect contaminants in astronomic
 al images. Once trained, our networks are able to detect various contamina
 nts such as cosmic rays, hot and bad pixels, persistence effects, satellit
 e or plane trails, residual fringe patterns, nebulosity, saturated pixels,
  diffraction spikes and tracking errors in images, encompassing a broad ra
 nge of ambient conditions (seeing), PSF sampling, detectors, optics and st
 ellar density. MaxiMask is performing semantic segmentation: it can output
  a probability map for each contaminant, assigning to each pixel its proba
 bility to belong to the given contaminant class, except for tracking error
 s where another convolutional neural network can assign the probability th
 at the entire focal plane is affected. Training and testing data have been
  gathered from real data originating from various modern CCD and near-infr
 ared cameras or simulated data. We show that MaxiMask achieves good perfor
 mance on test data and propose a prior modification technique based on Bay
 esian statistics to optimize its behaviour to any expected class proportio
 n in real data.
END:VEVENT
END:VCALENDAR